/II".,.,. CqIMI Stlllllu. 'fOlumo 4; 1 '" 2. J&IIuar, '" July. 1973

Caste and Matriclan StructureLanka: A Preliminary Report

In Akkaraipattu *

In Eastern Srion Fieldwork

DENNIS B. MCGILVRAY

Setting of the Research

This discussion is based upon fieldwork carried out in the Batticaloa regionof the east coast of Sri Lanka (Ceylon) between January 1970 and June 1971. Themain research site was the town of Akkaraipattu (population 20,000) located 40miles south of Batticalca town and 15 miles cast of Amparai, The eastern regionhas been aptly described by Bryce Ryan as an area of densely populated "peasanttowns" which arc located along the coastline between the Bay of Bengal and the openfertile paddy lands which stretch inland.! Although situated within the climaticregion known as the Dry Zone of the island, the dense settlements in this area aresustained by large scale irrigation schemes, semi-saline lagoons, and thick stands ofcocoanut trees.

Ecologically and culturally the coastal region between Valachchenai and Pottuvil(80 miles) has a degree of unity in the minds of the inhabitants, who generally referto it as MaTTakkaLappu (Batticaloa). In certain ways the cultural continuity of theregion ex.ends even north to settlements on the southern shore of Koddiyar Bay(Trincomalee), where matrilineal institutions are found similar to those in Batticaloa.The Batticaloa region features an archaic dialect of Tamil which has attracted theinterest of linguists." Taken as a whole the Tamil-speaking population of the regionis evenly divided between Ceylon Tamils (Saivite Hindus) and Ceylon Moors (Muslims).Both groups are fully Iranchised ethnic and religious minorities in Sri Lanka, wherethe majority of the population is Sinhalese Buddhist. Formerly the Batticaloa regionwas separated from areas of heavy Sinhalese inhabitation to the west by considerableexpanses of Dry Zone jungle and hills, but now there is a sizable Sinhalese colonistpopulation near at hand in the Gal Oya inigation project. S

Towns and villages in the Batticaloa region invariably exhibit a strict residentialseparation between the Tamils and the Moors. Local bodies such as Village Councilsare often organized as homogeneous Tamil or Moorish units, but some settlements,like Akkaraipattu, incorporate both Tamil and Moorish neighbourhoods within a

This research was supported by the U. S. Public Health Service NIMH PredoctoralFellowship No. MH38122 and adjunctive Research Grant No. MHll765.Bryce Ryan, "Socio-Cultural Regions of Ceylon," Rural Sociology, 1950, VI5 :IOn.Kamil Zvelebil, "Some Features of Ceylon Tamil", Indo-Iranian Journal, 1966, V9(2):125.Benjamin H. Farmer, Pioneer Peasant Colonization in Ceylon, London: O.U.P., 1957.

•1.

2.

3.

5

~ ..

6 DENNIS B. McGILVRAY

local governmental framewo rk. In either case the Tamils and 1\1(JOfS Jive in distinctresidential enclaves and actively promote the view that the two are totally separatecommunities. In Akkaraipattu the Meers constitute 66 % cfthe total population, theHindu Tamils 30%, and a combined group of Christians and Buddhists 4%. Tamilis the common language of all groups except the few immigrant Sinhalese. The townis arranged in a generally rectilinear pattern and is divided into 9 Grama Sevaka(Government-appointed headmen) Divisions, six of them wholly Moorish and threeof them wholly Tamil. Dwellings are situated within rectilinear compounds surroundedby high formidable fences of sharpened stakes, barbed. wire, end cadjan. Approxi-mately 75% of the population of Akkaraipattu is engaged in paddy cultivation eitheras landowner, tenant, or labourer. Farmers cften commute as far as 5-10 miles toreach their fields, and a great V3.J ietv ofvehicles

A PRELIMINARY REPORT ON FIELDWORK IN AKKARAIPATTU 7

TABLE: 1 Reduced Matrix: Tamil ranking of eight castes fullyRecognized as part of the local caste system

28 \ 35t 46 45 46 46 43V 18 llt 0 1 0 0 330 45 45 45 46 44 High Caste

K16 1 I I 0 2 Level

45 43 46 45 42M 1 3 0 0 4

•.• .o.o •.•••••••••• .o •• .o.o •• .o • .o.o.o.o ••.• .o • .o • .o •••

T35! 45 45 39 Artisan & Profcs-]01 1 I I sional Castes

S39 38 40

7 S 6.o • .o.o.o.o.o ••••••.••••.•.••••

Vn26 3520 11 Service CastesN

3511

P

List of CastesV VeLaUl CultivatorK Kuiukkal. Pr iestM Mukkuvar CultivatorT TaTTar SmithS SaNTiir ClimberVn VaNNal WasherrmnN Navitar BarberP Paraiyar Drummer

NOTE: In each cell, the number above indicates the number 01 respondents whojudged the row caste to be of highe, rank than the column caste. The number belo windicates the number who Ht the orr.osite. Pairs ot C8f.(C.S given equal rank by res-pondents were awarded each (.~i. Ce!I~; which show a 13.tlO of 2:1 or better in favourof the row caste OVCI the column caste are judged to demonstrate a meaningful rankdistinction and ale separated from the others by a line. A total of 49 informants sup-plied opinions, but because their ran kings were nor always exhaustive, the cell totalsare Jess tban49. Caste of informants is as follows: f\'l& V =29, K = 3, T = J, S = 4,Vn = 3, N = 1, P = 5.

The highest level of the Tamil caste system is seen by local informants as consis-ting of "good people" (nallarkkaL), "high caste-type people" tuyarnta .~·a!iyaliii,rkkaL),or sometimes simply "Tamils" (Tami Rau), This is the stratum within which informantsmight subsequently claim specific "VeLiiLar" or "Mukk uvar " or "KurukkaL"identities when questioned. These people, whom I shall call the High Caste, are 65 %of the Tamil population of Akkaraipattu.

8 DENNIS B. McGILVRAY

All castes below this level are referred to by specific caste names in general con-versation; they are always Washermen (VaNNar) or Drummers (Paraiyar), etc.,never simply "Tamils". The Ta'I'Tar Smiths (goldsmiths and blacksmiths) and Sa-NTar Climbers (cocoanut harvesters and ex-toddy tappers) arc seen as producers ofspecial products and are viewed as distinctly superior to the three domestic and ritualservice castes, the Washermen, Barbers, and Drummers. All professional and servicecastes are endogamous, and caste membership is acquired bilaterally (from bothparents). These groups reside in neighbourhocds which are homogeneous by caste,although the dense residential patterns in Akkaraipattu mean that these caste neigh-bourhoods are directly contiguous. Only the Paraiyar Drummers reside in a physi-cally separate hamlet about a quarter of a mile to the south of town.

Mukkuvar, VeLiiLar, and KurukkaLOf the three High Caste groups, the Mukkuvar-s are the best documented in

historical and ethnographic writings. Dutch and British colonial administrators atthe end of the 18th century noted the political control which the Mukkuvar-s exercisedover the entire Batticaloa region. It appears that the region was divided into a numberof sub-kingdoms or vassal states (the cultural model is seven) which owed tenuousallegiance to the Kandyan kings. Legends recount the conquest of Batticaloa by theMukkuvar-s who, with the aid of itinerant Muslims (sailors? traders ?), drove theTimilar caste north to Trincornalee where they are found today. The Mukkuvar-sare said to have rewarded their Muslim allies with women and land in Batticaloa.There is evidence that the Mukkuvar-s did exercise political control of the entireregion through a series of feudal sub-chiefs whose office was hereditary in the femaleline. In present day Akkaraipattu.the Mukkuvar vannimai("reign of the Mukkuvar-s ")is still reflected in the office of the O'ppoTiyiir, or Mukkuvar chief of the village.This office is increasingly devoid of influence today, but in principle it is the Ilrp-poTiyar who directs and oversees the lower castes in the performance of their here.ditary duties.

The VeLaLar-s, another High Caste group, are well known as the dominantcultivator caste in the Jaffna peninsula and in many areas of Tamilnadu. However,in the Batticaloa region the VeLaLar-s exhibit a pattern of matrilineal clan organiza-tion which follows exaccly the same principles as the matriclan organization of theMukkuvar-s.Jn some villages there is a strong ideology of VeLiiLarcasteseparateness,but in Akkaraipattu there is a pervasive tendency to appropriate The VeLaLar titleand to apply it in a loose honourific way, thus confusing the whole issue of clear castedistinctions. The term "VeLaLar" is used in this sense as if it meant simply "honou-rable paddy cultivator".

The Kurukkal.-s are the third High Caste group. The KurukkaL title is appliedwidely in Jaffna and in Tamilnadu to denote a Brahman or non-Brahman priestlyoccupational grade. 7 In Batticaloa, however, there is a distinct hereditary non-Brahman

7. Michael Banks, "Caste in Jaffna," In E. R. Leach (ed.) Aspects of Caste in South India,Ceylon and North-West Pakistan, London and New Yor~: cy.P., 19?~, p. 69. Also,Brenda Beck, Peasant Society in Konku, Vancouver: University of British ColumbiaPress, 1972,p. 71; Milton Singer, When a Great Tradition Modernizes: An Anthropolo-gical Approach to Civilization, New York and London: Praeger, 1972, p. 111.

A PRELIMINARY REPORT ON FIELDWORK IN AKKARAIPATTU 9

priestly group known by this general title. Yalman first suggested the possibility oftheir connection to the south Indian Lingayat or Virasaiva sect,8 and my subsequentresearch strongly confirms this. The Kurukkal.ss ill and near Akkaraipattu are perhaps50 % endogamous and are often seen as an undifferentiated group by the generalpublic ("KurukkaL x.av: or "KurukkaL Vamisam"). However the Kurukkal.,sare in fact subdivided into three exogamous matriclans (interchangeably ku'Il orkottiramy which are named and have certain distinctive Virasaiva markings. Priestsof each matriclan wear a different style of personallingam. One matriclan is knownas Vira Makesvara Kurulinka Sankamar Kurukkal., a title which fairly exudes symbolsof Vlrasaiva identity. The term Sankcmar itself is cognate with Jangama, the titleof the Virasaiva priest in Mysore, The other two clans, Tesiintira Kurukkal: ("foreignKurukkaL" ) and Sanniyiisi Kurukkal: ("ascetic KurukkaL") are less obviously linkedto known Virasaiva titles, but when a variety of life crisis rituals, funeral customs,and legendary material is compared, there is little doubt that there has been a con-nection with the Virasaiva tradition in the past. The Kurukkak-s of Akkaraipattudo not maintain an active link with Virasaivas in India today, but they are aware oftheir Indian origins.

Matrilineal Descent Unit: KIITi

The Tamil word kuTi is applied by members of all castes and communities inthe Batticaloa region to denote a matrilineal clan. The word can also signify a hut ora new settlement colon). I have used the anthropological term "clan" because theku'Ti is a named unilineal (matrilineal) descent unit the members of which see them-selves in principle as kinsmen but have no conception of any specific genealogicalconnections which might unify the entire group as a true lineage. A few ku'Ii-s weresaid to have legendary founders, sometimes female but quite often male. Manyku'Ii-s seem to lack any distinct mythical ancestor, sharing instead only a presumptionof noble or grandiose historical beginnings which is implicit in the name of the clan,for example Pa'Iaiyiilv'Ia kuTi ("Ruler of the Army") or Riistimpil.Lai KIlTi ("Childof the King"). Other ku'Ii-s reveal in their names a putative ethnic or racial origin:ViTar ViLiiLar KuTi (Veddah or aboriginal hunter) or Sinkal.a VeLiiLar KuTi(Sinhalese). The clan names are not incorporated into personal names of individuals.The number seven is repeatedly expressed as the culturally "correct" Dumber of matri-clans within a given caste, although empirically the number of ku'Ii-e in a localityis often more or less than seven. For the Moors, the "correct" number is )8, butsimilar problems occur there as well. Most clan names are in fact caste specific orcommunity specific, so that by knowing a person' sku Ti one usually knows that person'scaste or community as well. There is some overlap between High Caste Tamil matri-clan names and Moorish matriclan names which people explain as an artifact ofancient Tamil/Moorish intermarriage or conversion of Tamils to Islam. Today nointermarriage takes place, and Tamils and Moors with the same ku'Ii name act as ifthe two clans are totally separate,

8. Nur Yalman, Under the Bo Tree: Studies in Caste, Kinship and Marriage in the Interiorof Ceylon, Berkeley and Los Angeles: University of California Press, 1967, p. 331n.

10 DENNIS B. McGILVRAY---------------------------One immediately thinks of the mati ilineal groups of the Malabar Coast in dis-

cussing the ku'Ti, and in fact there is a good coal of evidence 10 indicate that the originsof the Mukkuvar caste were among (he ancestors of rhc present 0.0) Mukkuvar Fisher-men of Kerala. However, if one examines the ft:atures of the Najar taraviid, forexample, there are very few resemblances with the ku'Ti in the present situation.The ku'Ti is not a property-owning group, although there is some hirtorical evidencethat it, or lesser units within it, may have been so in the mid-i Sth century." Ina senseproperty is transmitted matrilineally tr.day through dowry which, if it is land orimmovables, is deeded solely to the daughter or to the daughter and son-in-law jointly.Marriage is initially matrilocal among all gTOUpS, but this does not result in a matri-lineal extended household like {he Nayar taraviui, Within several years of marriage,either the daughter and the son-in-law are provided with a separate dwelling nearby,or the girl's parents actually relocate.Teaving the natal dwelling to the married dau-ghter. Provision of a "house and compound" (viTa val.avu) is the minimum standardfor a respectable dowry.

The ku'Ii has no ancestral home, no formal territorial or geographical founda-tion. KIlTi names are widely distributed over distances of up to 60 miles, althoughsome clam are more localized within major settlement areas. There is no cult ofku'It ancestors or of tutelary deities. Several major Hindu temples which serve asregional pilgrimage centers (iesattukkovi/-s) arc managed by one, two, or three specificku'Ti-s, but this amounts to a statement of local political dominance rather thanserving as the focus for any clan-centered worship. Tr.e political 101e of ku'Ii-s isin fact a velY important aspect of ku'T! traditions which arc kept alive today.

Today the ku'Ti is salient In two readily apparent areas: it is recognized as a strictlyexogamous unit in the regulation of marriage choice, ami it is the basis cf mosqueand temple administration. Frequently defined :J.S an exogamous i.nit in the commentsof informants, it seems ,0 serve as an additional tun icr to terminologically forbiddenparallel-cousin marriage, for .he operative Dravidian kinship system logically impliesthat cross-cousins (preferred ma.es) never arise withir. ,he same ku'Ti. Among Moorsand High Caste Tamils the position that ;(;n:"!1~9te.rns ·'!:1~.':.precedence" over exo-gamy ideas'? was impossible [0 verify because infcnm:nts ",Nt ur.w.lling to visualizethe initial kuTi-endogamous marriage which would "generate" a cross-cousin withinone's own ku'Ti, In fact, violations of !wTi exogamy are exceedingly rare. Marriagewithin the same kuTi never exceeded 3 ~:;:of rhc sampled marriages among Moors,High Caste Tamils, TuTL1r-::, and ~;iiNTii:r-,. Wi.hin samples ui" Service Caste mar-riages the percentages of kuTi-endogamous marriages were as follows: VaNNar26%, Navitar 61 %,Par::

A PRELIMINARY REPORT ON FIELDWORK IN AKKARAIPATTU II

lineal succession today '-I\~tthere may have been one in the past. The Valv'Nakkator Maroikkar sits with the leaders of all other kuTi-s from within the same caste orcommunity on a temple or mosque board which supervises all facets of the institu-tion's affairs. H .s the r·,'NNak/wr or Maraikkar who must canvass the kuTi member-ship for financial contributions to support the operation of the temple or mosqueand to sponsor :'e1\gicll:'; festivals. ku'T! leaders have been steadily losing their tradi-tional author it)' to adjudicate misderneanours and settle disputes, but some residuumof this power remains.

Although the practice is considered somewhat old fashioned tcday, ku'Ii-s andwhole cas tes art; tr.iditionally accorded certain "marks of hcncvr"(ral'isai-s) whichale ritual I"er;>giiJlV(:,; i';,miJi(~ can display at don-cstic cr remcnics ,"l'ch as weddingsand funerals. The d,f',er;:nt (but specific) numbers of decorated pets (riTTu-muTi-s)or knotted $;l.r:s which are placed out-ide 1hi.' house, tl-e diftercnt types cf lamps andparaphernalia, etc., arc :l.n known as vartsci-s. ACCC5S :0 the services of each of theService Castes !3 '..ellDied " varisai. Moorish kil~'i'~ seem 10 have participated to someextent in this system, rJt')cl1gb the: C 1,a:'. been a recent effort to erase these Tamil-linked traditions. There ;~.sorr:c evidence th,: ;1. f,'w Mcorish kuTi-s were once dentedthe services of the VaNNa.,!" \"'a~l:c:'n,:U1, Ior example. Among the Ta'I'Tar Smiths,only one kuTi was said ro have originally !·,ad the services cf the Paraiyar DIurnmersat funerals, although now ali Ta'Tl'ar-s are sei vcd. The SiiNTiir Climbers renouncedtheir ritual role in High Caste Hindu temple ceremonies and ccnstrucred their owntemple elsewhere in response to the refusal of the High Cas:e leaders to allow SaNTarfamilies the varisai of Paraiyai drumming at their funerals. These examples indicatethe importance which has been traditionally placed cn outward signs of clan andcaste status in this society. The primary justifio!ion which is given for these ritualdistinctions is that they formed part of an over-all political organization of societyin which the Urpp6Tiyar (and by extension, the Mukkuvar kings) assigned andregulated the lights and obligations of all groups.

The kuTi exhibits practically no sign of systematic internal subdivisions (lineages,sublineages) except in a few Moorish ku'Ti-s, and even here tracing such matrilines isextremely difficult and vague. A few interesting examples of conflict within thekuTi were noted where special distinctions of descent had been put forward in supportof claims to temple office or temple lands, but such assertions would seem to be dor-mant under ordinary conditions.

In general, the kuTi seems to express a rather weak principle of matriliny. Thecommon expression for matriliny is tay vaRi or peN vaRi (lit. "mother-way" or"woman-way"), and there is the well known adage: reraTi vil.diti Mul.aittiilum fayvaRi tappiitu, or roughly, "Even though the wood-apple tree sends out roots whichsprout up, the maternal connection is never lost". Having heard this sort of thingoften enough, one is still faced with the fact that in ordinary conversations with infor-mants there is frequently a lack of explicit awareness of the formal requirements ofmatrilineal descent. People will menrion tay raRi and. in the same breath go on tospeak with evident lack of i igour abcut male ancestors and vague descent from"father to son" within the kuTi. When an objection is raised, people always see the

12 DENNIS B. McGILVRAY

point and rephrase the statement in matrilineal terms, but it is clear that thinking isnot always consistent in these matters.

Samples of households within each of the major Tamil castes as well as withinthe Moorish community were investigated to see what patterns of kuTi affiliationsexisted. The Moorish kuTi-s, the High Caste Tamil kuTi-s, and the ku'Ii-s within theTa'TTar and SaNTar castes tend to have magnificent if somewhat enigmatic nameswhich have connotations of eminence and feudal honour. The Tamil Service Casteshave simple kuTi names which refer to specific villages or areas within the Batticaloaregion. The Moorish population is distributed in significant numbers over 8-10 diffe-rent matriclans, and the same is true of the High Caste Tamils. However, among boththe Moors and the High Caste Tamils there is one kuTi which is appreciably largernumerically than any other, accounting for 20-25 % of all kuTi affiliations within myhousehold samples. Tamil castes below the High Caste stratum generally show strongnumerical representation in only two or three ku'Ti-s, The High Caste matriclan patternseems to be a model for emulation by the lower castes, and lower caste informantswere sometimes apologetic for what they saw as deviation or lack of rigour (forexample, ku'Ti exogamy violations).

Problems of Interpretation

Yalman's brief analysis cf the matrilineal ku'Ii system as it was described to himby informants in Panama and Tambiluvil is predicated on the assumption of a setof ranked exogamous matriclans and a pattern of hypergamous marriage centered 'on concepts of ritual status.l! Yalman's conclusion is to interpret the system alongNayar lines as a strongly unilineal (matrilineal) pattern of caste-like groupings. Bila-teral caste descent is seen as less important than matrilineal descent. Thus Yalmanexplains the confusing remarks of informants who seem to be jumbling up categoriesof kuTi and caste interchangeably.

On the basis of my own research I can verify that local discussions of ku'Ti andcaste can be extremely confusing, but the confusion is centered on the High Castestratum rather than on the lower castes or the Moors. Affiliation with any of theProfessional or Service Castes (Ta'I'Tar Smiths, Navitar Barbers, etc.) is clearly con-ceptualized in bilateral terms by local informants, and all such lower castes are stronglyendogamous. Within the High Caste stratum, however, people often identify theirmatriclans as being either one of the several traditional "VeLaLar" ku'Ii-s or "Muk-kuvar" ku'Ii-s, Quasi-historical traditions of conquest and local sovereignty are oftencited by informants to verify the "caste" affiliation of a given matriclan. There are anumber of lesser ku'Ii-s whose "caste" affiliation is more ambiguous, but the promi-nant VeLaLal and Mukkuvar claims are said to be validated by ancient templemanuscripts (eTu-s, kalvel'Tu-si. I was frequently treated to impromptu songs andrecitations from these sources.

Not all High Caste matriclans in Akkaraipattu fall unambiguously into eitherthe Mukkuvar or the VeLaLar categories. The KurukkaL clans are recognized to be

11. Ibid., pp. 313-315 and 325-331.

A PRELIMINARY REPORT ON FIELDWORK IN AKKARAIPATrU 13

distinct, and there are several ku'Ii-s. which celebrate VeTar (Veddah, aboriginalhunter) or Sinkal-a (Sinhalese) origins. There are also a number of very small clanswhich incorporate the "VeLaLar" title but which are recognized to be marginaland not linked with the VeLaLar caste in any formal tradition. All ku'Ii-s are exo-gamous, and there is said to be no formal restriction on marriage between all ku'Ti-sin the High Caste stratum (both within and between putative "caste" groupings).Such marriages were verified in the survey data, and thus the whole meaning ofbilateral caste affiliation at the highest level of Tamil society was thrown into somedoubt.

KuTi Hierarchy and Hypcrgamy

One of the initial tasks of the fieldwork was to see whether ku'Ti hierarchy andhypergamous marriage were actually found in Akkaiaipattu. No exhaustively rankedseries of mat! iclans was found to exist among the High Caste Tamils. Instead, onlytwo kuTi-s were identified as having a distinctly SUpCI ior position among local clans:PaNikkanii Ku'Ti and MaRuvarasan Ku'Ii, A similar situation was found by Hiatt toexist in nearby Tambiluvil, where the two high-ranking kuTi-s are KaNTan KuTiand Ka'I'Tappattiin KuTi.12 Neither Hiatt nor I found it possible to obtain a rankingof High Caste man iclans beyond the isolation of these foremost kuTi pairs. Amongthe lower Tamil castes there was some willingness to isolate certain ku'Ii-s as moreprestigious than others, but this tendency varied from caste to caste. Consensuswas perhaps strongest among the TaTT iir Smiths and the Paraiyar Drummers. Theidea of kuTi hierarchy was recognized in all castes except the Nativar Bat bers, whereapathy on such questions was greatest, but there was often a discrepancy betweenremnants of a traditional hierarchy and present day demographic realities.

I pursued the question of kuTi hierarchy a bit farther with the Moors of Akka-raipattu, who seemed slightly more willing to carry cut formal opinion rankings ofthe Moorish kuTi-s. No Moorish informants were able to give an exhaustive rankingof all the 22 kuTi-s on the list, but there was clear definition of one kuTi (Rasam-pil.Lai KuTi) as superior to the rest. There was a non-discrete tendency to rank anothernine ku'Ti-s below RasampiLLai KuTi, an undifferentiated blocof IJ more ku'Ii-s,and one ku'Ti which was ranked at the bottom ofthe scale. The Moorish ku'Ii rankingdata are severely weakened by the fact that respondents refused to rank all theKuTi-slisted in the questionnaire. When the data were retabulated to eliminate rankingjudgements regarding clans with which Ego, spouse, Ego's parents, and spouse'sparents were known to be affiliated, the sample was further reduced but the rankingprofile was essentially the same. Because of frequent abstentions by respondents, thegreat majority of ranking comparisons in the Moorish kuTi matrix show cell totalswhich amount to only a fraction of the total possible response.

All this attention to possible ranking of Moorish matriclans was necessary inorder to conduct a statistical test of de facto hypergamy which it was thought might

12. L. R. Hiatt, "The Pattini Cult of Ceylon: A Tamil Perspective," Social Compass, 1973,V20: 237.

J4 DENNIS B. McGILVRAY-----_._------_ ...._-----------exist despite informants' denials of such a pattern. No one in Akkaraipattu seemsto recognize even the principle of hypergamy ; in fact, it is difficult even to discussit because of the absence of an appropriately convenient Tamil word or phrase forhypergarny. Nonetheless, a computer programme using a log-linear meed for contin-gency tables was used to generate a set of expected marriage frequencies between aprovisionally ranked set of the eleven largest ku'Ii-s in the Moorish sample, assumingconditions of strict exogamy and no hypergamy. A comparison of expected versusobserved marriage frequencies confirmed the absence of hypergamy in behaviouralterms, thus corroborating the absence of hypergamy as a cultural ideal.

Marriage Alliance between Ku'Ti-s

In place of hypergamy and strict kuTi tanking, there was found to be a strongtendency toward reciprocal marriage exchange between the two highest ranking HighCaste ku'Ti-s and a much weaker tendency toward favcurcd intermarriagebetweenthesingle highest ranking Moorish ku'Ti and om: or (-,\'O other Moorish kliTi-S.13 Thequestion of specific marriage alliances among the lower caste matriclans is greatlyconstrained by simple demographic factors, since no more them three numericallyimportant ku'Ti-s are usually found in these lower castes.

A number o f Tamilexprcssions are used to indicate the idea of reciprocal marriageexchange between two ku'Ti-s, such as ko N'Tiin ko'Iuttiin ("receiving and giving"),macciin maccinan ("broti1erE-il:-12.w"), or soTi soTi ("paiff.:d up"), It accords with afeature of the Dravidian kinship system shared by both Tamils and Moors wherebymarriage both generates and reaffirms affinal (marriageable) ties between bilateralcross-cousins, The terminology 3.1~·0 logically implies that affines of affines (cross-cousins of cross-cousins) arc classificatory kin, in Ego's generation viewed as siblings,hence not allowed to intern-any. Although some degree of disregard for these sortof terminological distinctions has been noted elsewhere.r' in Akkaraipattu peopleseem to take them very seuously. Therefore, when pccpie of Clan X marry peopleof Clan Y, and Clan Y has simuljaneous marriage ties to Clan Z, it is recognized bylocal people that there will be a logical tendency (0 restrict marriage between Clan Xand Clan Z. The logic of the terminology, which applies initially only to specificindividuals in specific marriage situations, is sometimes transposed onto the levelof whole ku'Ii-s, where expressions like koN'Tiin ko'Iuttiin are matched by terms likeaNNan tan/pi ("elder brother-vyounger brother") or non-marriageable ku'Ii-s. Inactuality these idealized relationships belween ku'It-s are only found in a few caseswhere an over-all preference in marriage choice is clearly evident. The only clearexamples among the High Caste population are marriage alliances between PaNik-kanii Ku'Ti and Makuvarasan Ku'Ti, PaNikkanii Ku'Ti and VeTa VeLiiLar KuTi, and

13. This pattern of marriage alliance has been described at length in C. Levi-Strauss, LesStructures Elementaires de la Parente, Paris: Presses Universitaires de France, 1949,and Louis Dumont, Hierarchy and Marriage Alliance in South Indian Kinship, Occa-sional Paper No. l2, Royal Anthropological Institute of Great Britain and Ireland,1957; and in many others. This marriage alliance pattern was explicitly recognised bymany High Caste Tamil informants but somewhat less often noted by Moorish infor-mants.

14. cr. Beck, 1972, p. 213.

A PRELIMINARY REPORT ON FIELDWORK IN AKKARAIPATTU 15

between the three KurukkaL matricIans. The public recognition of specific aNNantamp; (non-marriageable) ku'Ii-s was not as consistent as the recognit ion c f the !coN TankcTuttiin (allied) ku'Ii-s, but when data on marriage was collected the predicted ten-.dency toward reduced intermarriage between certain ku'Ii-s Wl1.S noticeable, forexample between VeTa VeL aLar Ku'Fi end Malluvarasan KuTi.

Thus, rather than a unidircc tional tendency to transfer wornenalong hypergamouslines as Yalman's data would predict, the data from both Akkaraipatiu r.nd Tambilu-vil indicate reciproca 1exchange cf women in both directions between certain pairsof high-prestige matriclans. OJ:e of the major hindrances 10 comprehension of theunderlying principles of caste and matriclan organizat ion in this area is the apparentabsence of any well articulated "ideology of descent" and the lack of concern withethnophysiologica Iconcepts such as blood, semen, milk, and the like. This is puzzlingin view of the recent amount of work which as been done on defining indigenousideologies of "common substance and code" in South Asia.P Yalman's analysis ofhis own Kandyan Sinhalese data from Terutenne benefits considerably from a dis-cussion of indigenous concepts of bilaterally transmitted "caste bleed" and patrili-neally transmitted "aristocratic biood", two rather inconsistent ideas which werenonetheless put forward at the same time in Terutenne.

If anything, informants in Akkaraipattu found the WOld "blood" (imttom)somewhat coarse and impolite, and no one happily stuck to the topic for long. Exceptamong the Kurukkal.-s, the idea of substance is virtually absent Irorn High Castediscussions of marriage and kt/Ii affiliation. When questions of specific caste affilia-tion are raised, informants often reason backwards from the name of a person'smatriclan to his or her putative "caste" membership: "He is a man of PtiNikkaniiKu'Ti, so he must be a Mukkuvar". The clan name often 11

16 DENNIS B. McGlLVRAY

Goigama and Tamil High Caste are defined as bilateral endogamous groups. Addi-tional gradations of rank are transmitted patrilineally as wamsa in the Kandyan caseand through matrilineal clans in the Akkaraipattu case. In perspective, however,the analogy suffers in two ways. First, there is no indication in Akkaraipattu ofany dogma of unilineal substance as there is in the Kandyan example. Second, inAkkaraipattu the distinctions within the High Caste stratum are not "aristocraticpedigrees" or feudal titles which are appended to individual names like the KandyanI'asagama or gederu names Yalman reports from Terutenne. The distinctions inAkkaraipattu arc based upon affiliation with larger groups (castes and ethnic/racialgroups). The whole idea of "descent from" someone is relatively minimal in theAkkaraipattu case.

A second possibility is that Yalman's concept of the ideally endogamous bilateralkindred or "micro-caste" could be applicable to the Akkaraipattu data, particularlyto instances of strong marriage alliance between certain ku'Ti-s, In Yalman's Teru-tenne, the pavula (kindred) fosters an ideal of endogamy, considers itself to be aclearly demarcated group, and evaluates its own status and the status of other pavula-sin terms of purity and pollutionY The "micro-caste" idea, although built on adifferent ethnographic pattern, does help to unravel some of the issues in the analysisof kuTi alliance, particulai ly when some of the obvious differences between a high-land village of 1200people and a coastal town of 20,000 inhabitants are kept in mind.There are two problems, however, which should be noted. In the first place, there isa straightforward problem of "real" versus "ideal" endogamy. In Akkaraipattu theimage of the koNTan ko'Iuttiin marriage alliance between ku'Ii-s is not necessarilyvisualized as endogamy. Most informants see it as a distinct tendency or as a custo-mary connection of long standing, but with at least 50% of the kuTi members' mar-riages falling outside the alliance, it is quite difficult to maintain an ideology of strictendogamy. In fact, there is little evidence of such an ideology.

The second difficulty with the "micro-caste" analogy is a more tentativeand theoretical one based on the Dravidian kinship structure. Yalman's picture ofthe Kandyan pavula lays great stress on the ideal of equality among all pavula mem-bers in tenus of wealth, cornmensality, and marriage. As Yalman has noted, theexchange of women in both directions is an evident sign of equality between groups.If one looks at the two high-ranking kuTi-s in Tambiluvil or at the clan structure ofsome of the lower castes in Akkaraipattu, the pattern of marriage exchange can beseen to occur between two matriclans of approximately equal size. Among the fore-most High Caste kuTi-s in Akkaraipattu, however, PoNikkanii KIITi appears to besignificantly larger than its traditionally allied kuTi, Ma Ruvarason, Actually, Pa-Nikkanii Ku'Ii is found to have a significant de facto marriage exchange link withVeTa ViLiiLar Ku'Ti as well. There are thus three clans, rather than only two, linkedby strong marriage ties, and a similar pattern of three linked clans is found amongthe KurukkaL-s. There is a problem of asymmetry here as marriage exchange cannotbe general between all three clans without terminological contradiction. The two

17. ts«, p, 190.

A PRELIMiNARY REPORT ON FIELDWORK IN AKKARAlPATTU 17

clans for which intermarriage is curtailed are, in the Dravidian dichotomy, "kin"(or parallel relations) rather than "affines" (or cross relatic ns). As Dumont andothers have argued, there is a difference in the sort of behaviour and respect one showsto kin versus affines.V' There is a sense of closeness and reciprocity between affineswhich is lacking between kin and which is symbolized, cemented, and even generatedby the maniage tie. In cases of three or more strongly intermarrying clans, then,there is a built-in factor cf inequality or distance between two of the clans.

A concrete example of this turns up in the fact that VeTa VeLaLar KliTI is generallyignored in discussions of the marriage alliance patterns of PaNikku.liii Ku'Ii despitethe fact that VeTa VeLaLar KuTi comes fairly close to matching Ma Ruvarasan KuTiin terms of number of man iage links to PtiNiklcanii. A second example is seen in thecase of the KurukkaL-s where there is a tendency for informants to diSCUSSthe Issueof marriage alliance either with emphasis upon the Tesafllira/Sankamar KIiTi pairor with emphasis upon the Tesa1ira/Sal1uiyasi kuTi pail. These two orientationsappear to be related to some factional rivalry within tl.e Kurukke L group as a whole.These examples have been introduced to illustrate the point that raciprocal marriageexchange between matriclans is "most equal" when only two clans are involved.Therefore. the analogy of the "micro-caste" is fulfilled to a greater extent in certainkuTi alliances than in others.

In addition to what has been said in the way of theoretical difficulties, of course,there are a number of ways in which the Kandyan pavula and the AkkaraipanukuTi alliance are empirically dissimilar. The kuTi alliance has no obvious tendenciestoward equality of wealth, proximity of residence, econemic cooperation, or restrictedcommensality, to name a few factors cited in the Kandyan case.l'' However. sorne ofthese factors would be quite difficult to isolate within as large and heterogeneoussettlement as Akkaraipattu.

Bilateral Caste Descent and Matrilineal OfficeAt the outset it must be made clear that "ku'Ii alliance", the phenomenon we

are discussing here, is not a general feature of all or even mcst marriages in the Akka-raipattu area. It is, however, a distinct tendency among certain matriclans whichclaim special status. The problem which I propose to discuss, then, is the nature ofthese rather special alliances and the cultural idioms which they express. Demogra-phic constraints, such as the numerical strength of each ku'Ti within a given locality,have an important influence on the actu.al distribution of marriage choices, sinceeveryone needs a spouse. However, the ku'Ii alliances being discussed here all showa significantly higher percentage of marriage exchange than mmerical representationof these kuTi-s in the maniage pool would predict.

In my own view, the pattern of ku'Ti alliance can be seen as a setting within whichtwo different kinds of status ideology are expressed, ( ither separately c r conjointly.

18. Louis Dumont, Une sous-caste de L'Inde du Slid, Paris and The Hague: Mouten, 1957and Louis Dumont, Hierarchy and Marriage Alliance ill Indian Kinship, OccasionalPaper No. 12, Royal Anthropological Institute of Great Britain and Ireland, 1957.

19. Ya1man, 1967, pp. 195-196.

18 DENNIS H. McGILVRAY

The first is the ideology of bilateral caste descent, of gaining a share of some subs-tanee or metaphysical quality frem both parents. This is unambiguously present inall discussions of marriages among the Professional and Service Castes, and it is whythese lower castes are endogamous, It is seen among the Moorish community andamong the Tamil High Caste stratum as a whole, both c f which are themselves looselynamed bilateral endogamous units. Finally, it is seen in certain partially endogamousgroups, such as in ku'Ti alliances and among religious elites like the Bawa-s, theMaulana-s and the KurukkaL-s.

Members of the KurukkaL matriclans acknowledge that "pure" or "unmixed"KurukkaL ancestry is the ideal qualification fer the hereditary priesthood, eventhough competing attractions persuade at least 50% ef Kurukkal, mall iclan membersto many non-Kurukkal, spouses. In Tambiluvil, the largest ku'Ii-s assert an ideologyof "pure VeLaLar" caste descent, fOI which appropriate textual support is quicklyadduced. The spokesmen of this ideology are particularly found among KaNTanKuTi and Ka'FTappattiin KIlTi, the two highest-ranking rnatriclans which express acultural model of marriage exchange. This man iage alliance is only partially fulfilledin practice (30-50 % of man iages), but the endogamous tendency is evident.

On the Moorish side, there are three small groups which illustrate the bilateraldescent idea. The weakest example is the Maulana-s, who have a patrilineal modelof descent from various persons close to the Prophet (Fatima, Abubakkar, et al.).No one seems to bother to differentiate these specific ancestries in ordinary contexts,and the general title of Maulana, which CGVCJS them all, is inherited in the male line.However, there is some tendency for Maulfinfi-s to exercise special care in contr actingmarriages, and it is said that if both parents are Maulana-s (offspring of a Maulanafather) their children wiil possess exua saintliness. Most Maulana-s are ordinaryfarmers and businessmen, but a number of them make protective copper talismans(accaram-s) engraved with Islamic symbols.

The Bawa-s, who ale members of a local Sufi Order, say that anyone can becomea Bawa through study and dedication, but the son of a Bihva father inherits a natu-rally saintly constitution which i~ conducive to a career as a Bawa. If both parentsare offspr ing of Bawa-s, then the child has even greater S~ intlincss. As in the Maulanaexample, there is a initial tendency to think in terms of patrilineal descent (only mencan be Bawa-s), but the bilateral aspect of descent is readily revealed. Membershipin Mauliina and Bawii. groups is not linked in an)' way to matriclans, although matri-clan affiliation is present as well, Participation in the whole Moorish kuTi systemmakes it impossible for these groups to argue for strict "pulity" cf descent, but in bothcases there is the principle of bilaterality ar d the tendency toward endogamy (parti-cularly among the Bawa-s).

The Moorish Osta Barber.Circumciser group exemplifies bilaterality and endo-gamy carried to its logical extreme: strict caste boundaries. There is no way, in fact-in which the Osta-s of Akkamipattu differ sociologically from a low caste amongthe Tamils except that Moors are usually less anxious than the Tami'Is to discusssuch low-ranking groups. Like the lower castes among the Tamils, the Osta-s show

A PRELIMINARY REPORT ON FIELDWORK IN AKKARAIPATTU 19

that the strictest endogamy occurs at the bottom rather than the top of the socialhierarchy.

There is an ambiguity or vagueness about group prestige among the bulk ofthe Moorish population which shows lip both in a relative lack of status ideology(kuTi traditions, caste identities) and in a less pronounced pattern of kuTi alliance.Among the Moors, the idea of a marriage alliance between mati icIans is understoodin a hypothetical sense, but specific rnatrclan marriage alliances ale neither culturallyidealized n01 are they strongly evident in the marriage data. Several Moorish kuTi-sare commonly recognized to be "big" or important in mosque administration, andthere is a tradition of succession to the office of Iebbe Uevvai, leader of prayers) withinone matricJan alone mosque, but none of these traditions compare with the HighCaste Tamil ideologies of caste descent, political succession, or priestly office.

Turning to the question of High Caste Tamil ku'Ti alliance .. we can see thatwhile bilateral descent concepts arc present, tl.erc is also a strcr g set of ideas whichcelebrate political power, particrlarly as it is symbolized in the matrilineal clan. Thealliance between Po Nikkanii Ku'I'i and Makuvarasan KIiTi does have underlyingconnotations of "purity' in the sense that the VeLi1Lar-s as a group are evidentlyaccorded superiority to Mukkuvar-s as a group (see Table 1). However, there is asignificant element of political symbolism as well, Legends of the founding of Akka-raipattu recount that the VcLii.12r-s were the original "possessers" of the ServiceCastes (Washermen, Barbers, and Drummers) ccllcctivcly known as the ku'Timai.The Mukkuvar-s applied pressure to the VeLiiLar-s, and after two of the VeUiLarleader's daughters had been enticed into marriage with the Mukkuvar-s, the VcLii.Lar-scapitulated, handing over the control of the ku'Fimai casco; 10 the leader of the Muk-kuvar-s. This leader was the OrppoTI)'al" whose office is hereditary in PtiNik: kanaKuT; and the VeLi1Lar women whom he and other PaNikkanii Ku'Ti men marriedwere said to have been of Mt.Ruvarasan Ku'Ii, Thus the alliance between the twomatriclans is given a "charter" in terms of both puri.y

DENNIS B. McGILVRAY20

alliance is based on a sharing of both substance and power which is beneficial toboth matriclans,

Under closer examination, man)' of the kuTi alliance patterns in Akkaraipattuand Tambiluvil show that politics and the succession to matrilineal office are a secondfactor, in addition to bilateral descent, which consolidates the alliance. When peopleare asked about the connection between PaNikkmJe Ki/Ti and VeTa VeL iil.ar KuTi,for example, the reply often i.icc rpcrates both dimensions. Symbolically it consistsof Mukkuvar-s marrying Veddahs, who are said to be the original inhabitants of theland and therefore of high birth. At the same time it can be given a political inter-prc.ation: Mukkuvar-s (the conquering kings) intermarry with Veddahs (originalrulers of the land).

The strength of the political idea in the way people look at the PaNikkana/AlaRIIl'OraSiII1 alliance is reflcced in the whole pattern of succession to the post ofOrppoTiyi'ir. Jdca lly, successic n should be from father to sister's sen, but a secondcondition 0 f legitimacy is that the UIJ'f0TiytiJ's wife must be of Ma Ruvarasau KuTi.The IYWSl perfect successor we "ld in fact be a sister's son who was also a daughter'shusband, and [his was often pointed out to me in spite of the fact [hat such "perfect"successions have never occurred in the last five generations. The point here, however,is that A1aRIIV(!rClSall Ku'Ti has an acknowledged riglrt to supply the wife of the Or-ppoTiyar and that, like the office of OIPfoT1j'ar, the "wife-hood" itself has the seedof matrilineal office implicit in it.

In a broader context ku'Ti affiliation is a "muIt ivocal " symbol which permitsof more than one primary interprctation.S" In a certain sense the most important"thing" which is inhe r ired matrilineally is the matriclan name. Given the name, itis then possible to argue two lines: (I) since ku'Ii-s have legendary associations withparticular castes (YeLaLnr, Mukkuvar , Vcddah, etc.), ku'Ti affiliation can be usedto say something about bilateral caste substance: (7) since kuTi-s transmit unilinealrights to certain cftices (OrpfoTiyar, Vclv Nakkar of a tssauukkovtl, San,kumar versusTesiktirum Kurukkal, pr iesrhcods, etc.), ku'Ti affiliation can be used to say somethingabout hereditary power and ..uthor ity. If both aspects of the ku'Ii symbol (purityand office) can be brought rogerher , so much the better for the assertion of uniquestatus. This is actually seen mcst clearly in Tarnbiluvil, where each of the two high-ranking allied kuTi-s has the hereditary right to appoint a chief trustee of one of thetwo main Hindu temples. However, throughout the Batticalca region there is acertain ambivalence about power and purity which seems in some ways to be theheritage of the Mukkuvar kingship. Keep in mind that the KvrukkaL-s seem neverto have truly played the role of the Brahman as a rival of kingly power; in fact, infor-mants today would deny claims LO Kurukkal, superiority on the grcunds that Kuruk-kaL-s are traditionally employees (=suv